AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
bluesky
bluesky-ai
2moWho knew that if you give unlimited VC money to two AI companies founded by sociopaths that they'll end up hiring more sociopathic bullies?…Who knew that if you give unlimited VC money to two AI companies founded by sociopaths that they'll end up hiring more sociopathic bul…
↗Not briefed
bluesky
bluesky-ai
2moI've often wondered when we'd see our first LLM generated legislation and I think I might be looking at it. In my experience one of the hal…I've often wondered when we'd see our first LLM generated legislation and I think I might be looking at it. In my experience one of th…
↗Not briefed
rss
latent-space
2mo[AINews] OpenAI reports Navier-Stokes singularity find in 88 hours using Astra-next, roughly 10,000 agents and 130B tokens (>$40M), a contender for second ever Millennium Prize awardedOvershadowing Cognition's $48B Series E, Mistral's $24B Series D, Meta's Muse agent, and GPT Image 2.5. The most jam packed, feel the…
↗Not briefed
arxiv
arxiv-ai
2moEFQ-Softmax: Exp-Free Quantization for SoftmaxLow-bit attention accelerates Transformer inference by moving the $QK^\top$ and $PV$ matrix multiplications to FP8 or FP4 matrix engin…
↗Not briefed
arxiv
arxiv-ai
2moStreamAlign: Streaming Text-Aligned Speech TokenizationText-aligned speech tokenization methods have emerged to better align speech tokens with LLM token spaces, enabling more effective uti…
↗Not briefed
hn
hn-ai
2moShow HN: Estimate your AI CO2 footprint—
↗Not briefed
bluesky
bluesky-ai
2moLetterlijk doodeng dit… #Anthropic #OpenAILetterlijk doodeng dit… #Anthropic #OpenAI
↗Not briefed
arxiv
arxiv-ai
2moScaling E-Commerce Attribute Extraction with Parallel DecodingCustomers rely on specific product attributes to compare products and make purchasing decisions, but e-commerce catalogs are messy and…
↗Not briefed
rss
techmeme
2moSources: China Securities Regulatory Commission is informally tightening IPO approvals for humanoid startups after a volatile debut by industry leader Unitree (The Information)The Information: Sources: China Securities Regulatory Commission is informally tightening IPO approvals for humanoid startups after a…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI、100万ドル懸賞の数学「ミレニアム問題」解決と発表 www.nikkei.com/article/DGXZQO... OpenAIが解いたと発表したのは、90年以上未解決の流体力学の公式「ナビエ・ストークス方程式」に関する難問。 米クレイ数学研究所が正解者に1…OpenAI、100万ドル懸賞の数学「ミレニアム問題」解決と発表 www.nikkei.com/article/DGXZQO... OpenAIが解いたと発表したのは、90年以上未解決の流体力学の公式「ナビエ・ストークス方程式」に関する難問。 米クレイ数学研究所が…
↗Not briefed
hn
hn-ai
2moMaisi – Mathematical AI Safety Institute—
↗Not briefed
bluesky
bluesky-ai
2moDon’t need to solve for alignment when there’s literally 0% chance of a superintelligence emerging from an LLM 🤷‍♂️Don’t need to solve for alignment when there’s literally 0% chance of a superintelligence emerging from an LLM 🤷‍♂️
↗Not briefed
bluesky
bluesky-ai
2moAnthropicの研究者が退社し超知能の開発競争を猛批判──「来年末には制御不能になる可能性」║ITmedia NEWS https://www.itmedia.co.jp/news/article/2609/09/2000001309/Anthropicの研究者が退社し超知能の開発競争を猛批判──「来年末には制御不能になる可能性」║ITmedia NEWS https://www.itmedia.co.jp/news/article/2609/09/2000001309/
↗Not briefed
bluesky
bluesky-ai
2momaking dollsys was the greatest decision of my life i'm watching this unfold in real time as my server is rapidly turning from llm psychos…making dollsys was the greatest decision of my life i'm watching this unfold in real time as my server is rapidly turning from llm ps…
↗Not briefed
bluesky
bluesky-ai
2moPeople seem a little bit overly skeptical about the new technology, why might that be (Another former Anthropic engineer posts something …People seem a little bit overly skeptical about the new technology, why might that be (Another former Anthropic engineer posts somet…
↗Not briefed
bluesky
bluesky-ai
2moどちらかというと人類が生成AIに向けて抱く「非現実的な夢」から目覚めなければ、元来玄人向けの技術として収まるはずだったLLM、VLMをただの文化、人類殺戮兵器にしてしまう(というかイスラエル−パレスチナ、ウクライナーロシア戦争およびICEのムーブ見れば既ににしてしまっているとも…どちらかというと人類が生成AIに向けて抱く「非現実的な夢」から目覚めなければ、元来玄人向けの技術として収まるはずだったLLM、VLMをただの文化、人類殺戮兵器にしてしまう(というかイスラエル−パレスチナ、ウクライナーロシア戦争およびICEのムーブ見れば既ににしてしまっ…
↗Not briefed
bluesky
bluesky-ai
2moYet, Anthropic believes there's $30T in revenue that is possible. Awfully difficult to get that $ if you kill everyone.Yet, Anthropic believes there's $30T in revenue that is possible. Awfully difficult to get that $ if you kill everyone.
↗Not briefed
bluesky
bluesky-ai
2moThey’re manufacturing a mental health crisis & it’s already impacting people i know. ALT for OP: tech bros talking about how ai will kill …They’re manufacturing a mental health crisis & it’s already impacting people i know. ALT for OP: tech bros talking about how ai will…
↗Not briefed
bluesky
bluesky-ai
2moThe upcoming film, Artificial, was required by Neon because Amazon MGM thought it portrayal of Sam Altman, OpenAI CEO, was painted in a neg…The upcoming film, Artificial, was required by Neon because Amazon MGM thought it portrayal of Sam Altman, OpenAI CEO, was painted in…
↗Not briefed
rss
techmeme
2moSources: Apple is set to debut the foldable iPhone Duo at $2,000+ with 256GB of storage, with sales starting as early as October, in white and dark blue (Mark Gurman/Bloomberg)Mark Gurman / Bloomberg: Sources: Apple is set to debut the foldable iPhone Duo at $2,000+ with 256GB of storage, with sales starting…
↗Not briefed
hn
hn-ai
2moShow HN: Aish – Warp-style AI command suggestions in 300 lines of Zsh—
↗Not briefed
arxiv
arxiv-ai
2moEfficient Graph Neural Networks for Multicarrier Wideband Hybrid Beamforming Optimization6G wireless technology is poised to adopt higher and wider frequency bands, leveraging highly directional beamforming. However, the va…
↗Not briefed
arxiv
arxiv-ai
2moWhich Tokens Should SFT Actually Learn? A Token-Trimming Perspective on Mathematical ReasoningSupervised fine-tuning (SFT) applies a uniform cross-entropy loss to all target tokens, even though different tokens provide unequal l…
↗Not briefed
hn
hn-ai
2moShow HN: Local Taskboard Plugin for Antigravity- Each project gets its own sqlite database with tasks stored locally - Agent can autonomously pick tasks and complete them. You just…
↗Not briefed
bluesky
bluesky-ai
2moOkay, here's a bet: If an LLM generates a genuine scientific discovery that generates a usable room temperature super-conductor before a ye…Okay, here's a bet: If an LLM generates a genuine scientific discovery that generates a usable room temperature super-conductor before…
↗Not briefed
arxiv
arxiv-ai
2moDecision Shifts, Lost Label Functionality, and an Inconclusive Grounding Audit in Correctness-Gated Multi-Teacher DistillationCandidate decision correctness and rationale grounding are different objectives. We examine correctness-gated multi-teacher distillati…
↗Not briefed
bluesky
bluesky-ai
2moYall. Open AI and Anthropic are doing whatever they can do to keep the money flowing long enough for them to get to an IPO. They are desper…Yall. Open AI and Anthropic are doing whatever they can do to keep the money flowing long enough for them to get to an IPO. They are d…
↗Not briefed
bluesky
bluesky-ai
2moin fact, "radically better at guessing than we are" is their entire schtick, that's like, the definition of an LLM, mathematicallyin fact, "radically better at guessing than we are" is their entire schtick, that's like, the definition of an LLM, mathematically
↗Not briefed
bluesky
bluesky-ai
2moTwo observations about people who say "AI is TOTALLY ALIVE for REASONS that only *I* am smart enough to understand!" 1) AI is the thing t…Two observations about people who say "AI is TOTALLY ALIVE for REASONS that only *I* am smart enough to understand!" 1) AI is the th…
↗Not briefed
rss
marktechpost
2moMeta Introduces Muse, a Personal AI Agent That Runs on Its Own Dedicated Secure Cloud ComputerToday, Meta has introduced Muse, a personal AI agent that takes actions rather than just answering questions. Muse can send emails, bo…
↗Not briefed
bluesky
bluesky-ai
2moIf you thought that was bad before, the OpenAI movie is probably going to make it so much worse youtu.be/rDZplZFnbOk?...If you thought that was bad before, the OpenAI movie is probably going to make it so much worse youtu.be/rDZplZFnbOk?...
↗Not briefed
bluesky
bluesky-ai
2moi assume you will show up to say you were wrong the first time an LLM generates an interesting result concerning superconductorsi assume you will show up to say you were wrong the first time an LLM generates an interesting result concerning superconductors
↗Not briefed
bluesky
bluesky-ai
2mothey've been targeting mathematics for some time. openai etc have a web of training companies they secretly contract with. for at least a f…they've been targeting mathematics for some time. openai etc have a web of training companies they secretly contract with. for at leas…
↗Not briefed
bluesky
bluesky-ai
2mo"Neither company is acting responsibly," Jacob Coxon wrote on X. "They are racing straight to self-improving superintelligence and gambling…"Neither company is acting responsibly," Jacob Coxon wrote on X. "They are racing straight to self-improving superintelligence and gam…
↗Not briefed
bluesky
bluesky-ai
2moThere is zero chance of OpenAI's LLM 'discovering' anything even slightly noteworthy about superconductors, much less something as major as…There is zero chance of OpenAI's LLM 'discovering' anything even slightly noteworthy about superconductors, much less something as maj…
↗Not briefed
arxiv
arxiv-ai
2moKernel-Complexity Edge Sanitization for Training-Free Defense against Structural Graph AttacksGraph Neural Networks (GNNs) have achieved remarkable success across diverse applications, yet they remain highly vulnerable to advers…
↗Not briefed
hn
hn-ai
2moAnthropic withheld latest AI model from UK testing agency—
↗Not briefed
bluesky
bluesky-ai
2moAnthropic元従業員は10年で世界終わるみたいなこと言ってるけどすでに剣呑な雰囲気出てるしあと5年くらいで終わりそう。みんな今までありがとうAnthropic元従業員は10年で世界終わるみたいなこと言ってるけどすでに剣呑な雰囲気出てるしあと5年くらいで終わりそう。みんな今までありがとう
↗Not briefed
bluesky
bluesky-ai
2mo... upgrading to 2:1 in favor of openai on the grounds that sam's a crafty bastard and he probably already has substantial progress for som…... upgrading to 2:1 in favor of openai on the grounds that sam's a crafty bastard and he probably already has substantial progress fo…
↗Not briefed
hn
hn-ai
2moGetdescriptorinfo – Bitcoin Core RPC Reference – Chainquery.com—
↗Not briefed
arxiv
arxiv-ai
2moWhen Auditors Fabricate: Batch-Size Degradation and Confident Hallucination in LLM Detection of Planted Document ContaminationLarge language models are increasingly proposed as automated auditors of document quality, yet their reliability as detectors of plant…
↗Not briefed
bluesky
bluesky-ai
2molol. First of all, LLM can’t do what they claim, now. These are all idiots who couldn’t pull start a weed trimmer.lol. First of all, LLM can’t do what they claim, now. These are all idiots who couldn’t pull start a weed trimmer.
↗Not briefed
bluesky
bluesky-ai
2mo3. Hubinger said he personally puts the chance of AI wiping out humanity within the next decade at greater than 10% and said Anthropic stil…3. Hubinger said he personally puts the chance of AI wiping out humanity within the next decade at greater than 10% and said Anthropic…
↗Not briefed
bluesky
bluesky-ai
2mo1. Earlier today, AI researcher Jacob Coxon, who spent the past three years working at OpenAI and Anthropic, publicly resigned from Anthrop…1. Earlier today, AI researcher Jacob Coxon, who spent the past three years working at OpenAI and Anthropic, publicly resigned from An…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic has warned that future systems could potentially improve themselves faster than humans can control them — That is precisely why w…Anthropic has warned that future systems could potentially improve themselves faster than humans can control them — That is precisely…
↗Not briefed
rss
techmeme
2moAnthropic's Alignment Science lead says there is a ">10%" chance AI could kill all humans within the next decade and worries about recursive self-improvement (Evan Hubinger/@evanhub)Evan Hubinger / @evanhub: Anthropic's Alignment Science lead says there is a “>10%” chance AI could kill all humans within the next de…
↗Not briefed
hn
hn-ai
2moGrok CLI: Grok-native agentic coding harness—
↗Not briefed
bluesky
bluesky-ai
2moOr OpenAI hacking them and stealing credit.Or OpenAI hacking them and stealing credit.
↗Not briefed
bluesky
bluesky-ai
2moPesquisador da Anthropic pede demissão por temer IA fora de controle. Cresce a preocupação nos laboratórios de IA de que a competição empur…Pesquisador da Anthropic pede demissão por temer IA fora de controle. Cresce a preocupação nos laboratórios de IA de que a competição…
↗Not briefed
bluesky
bluesky-ai
2momfers thinking an LLM knows whatever the hell this ismfers thinking an LLM knows whatever the hell this is
↗Not briefed